'commit'

Controller/DouYinController.py

@@ -7,357 +7,263 @@ import asyncio
 from datetime import datetime
 from typing import List, Optional
 
-from fastapi import APIRouter, HTTPException, BackgroundTasks
-from fastapi.responses import StreamingResponse
-from pydantic import BaseModel
-import pymysql
-
-# Import custom modules
-from Config.Config import OBS_CLOUD_PREFIX, OBS_BUCKET, OBS_TMP_PREFIX, DORIS_HOST, DORIS_PORT, DORIS_USER, DORIS_PWD, DORIS_DATABASE, OBS_SERVER
-from Util.DouYinDownloader import DouYinDownloader
-from Util.ObsUtil import ObsUploader
-from Util.ASRClient import ASRClient
-from Util.LlmUtil import get_llm_response
-
-# Logger setup
-logger = logging.getLogger(__name__)
-
-router = APIRouter()
-
-# Database connection
-def get_db_connection():
-    return pymysql.connect(
-        host=DORIS_HOST,
-        port=DORIS_PORT,
-        user=DORIS_USER,
-        password=DORIS_PWD,
-        database=DORIS_DATABASE,
-        charset='utf8mb4',
-        cursorclass=pymysql.cursors.DictCursor
-    )
-
-class ParseRequest(BaseModel):
-    text: str
-
-class SummaryRequest(BaseModel):
-    ids: List[str] = []
-
-def update_status(id, status, error_msg=None):
-    try:
-        conn = get_db_connection()
-        cursor = conn.cursor()
-        if error_msg:
-            sql = "UPDATE t_douyin_record SET status=%s, error_msg=%s WHERE id=%s"
-            cursor.execute(sql, (status, error_msg, id))
-        else:
-            sql = "UPDATE t_douyin_record SET status=%s WHERE id=%s"
-            cursor.execute(sql, (status, id))
-        conn.commit()
-        conn.close()
-    except Exception as e:
-        logger.error(f"DB Error update_status: {e}")
-
-def update_record(id, title, obs_url, transcript, status):
-    try:
-        # Truncate title to 100 chars to fit DB schema (approx 400 bytes max for utf8mb4)
-        if title and len(title) > 100:
-            title = title[:100] + "..."
-            
-        conn = get_db_connection()
-        cursor = conn.cursor()
-        sql = """
-            UPDATE t_douyin_record 
-            SET video_name=%s, obs_url=%s, transcript=%s, status=%s
-            WHERE id=%s
-        """
-        cursor.execute(sql, (title, obs_url, transcript, status, id))
-        conn.commit()
-        conn.close()
-    except Exception as e:
-        logger.error(f"DB Error update_record: {e}")
-
-async def process_video_task(url: str, request_id: str, share_text: str = ""):
-    logger.info(f"Processing task {request_id}")
-    
-    # 1. Update status
-    await asyncio.to_thread(update_status, request_id, "PROCESSING")
-
-    temp_dir = os.path.abspath(f"temp_{request_id}")
-    try:
-        if not os.path.exists(temp_dir):
-            os.makedirs(temp_dir)
-            
-        # 2. Parse & Download
-        downloader = DouYinDownloader()
-        # url is passed directly now
-        if not url:
-            raise Exception("No valid URL found")
-            
-        logger.info(f"Downloading from {url}")
-        # Run download in thread to avoid blocking main loop
-        local_video_path, title = await asyncio.to_thread(downloader.download_video, url, temp_dir)
-        
-        # Title handling strategy:
-        # Priority 1: Extracted from share text (if available and valid)
-        # Priority 2: Extracted from video download (often "Unknown Title")
-        # Priority 3: Generated by LLM (done later)
-        
-        extracted_title = downloader.extract_title_from_text(share_text)
-        logger.info(f"Extracted title from text: {extracted_title}")
-        
-        # If we have a valid extracted title, use it. 
-        # But if we don't have a title yet (or it's Unknown), we definitely want to use extracted_title.
-        # Even if we have a title from yt-dlp, if it's just "Unknown Title", we prefer extracted one.
-        if extracted_title and extracted_title != "Unknown Title":
-            title = extracted_title
-        elif not title:
-            title = "Unknown Title"
-            
-        if not local_video_path or not os.path.exists(local_video_path):
-             raise Exception("Download failed")
-             
-        # 3. Upload Video to OBS (Long term storage)
-        logger.info("Uploading video to OBS...")
-        uploader = ObsUploader()
-        video_filename = os.path.basename(local_video_path)
-        obs_video_key = f"{OBS_CLOUD_PREFIX}/DouYin/{video_filename}"
-        
-        success, _ = await asyncio.to_thread(uploader.upload_file, obs_video_key, local_video_path, OBS_BUCKET)
-        if not success:
-            raise Exception("OBS Upload failed")
-            
-        # Construct public URL (Assuming standard OBS pattern or Config logic)
-        obs_url = f"https://{OBS_BUCKET}.{OBS_SERVER}/{obs_video_key}"
-            
-        # 4. Convert to MP3
-        logger.info("Converting to MP3...")
-        mp3_path = os.path.splitext(local_video_path)[0] + ".mp3"
-        cmd = [
-            "ffmpeg", "-y", "-i", local_video_path,
-            "-acodec", "libmp3lame", "-ar", "16000", "-ac", "1", "-q:a", "2",
-            mp3_path
-        ]
-        # Run ffmpeg in thread
-        result = await asyncio.to_thread(subprocess.run, cmd, stdout=subprocess.DEVNULL, stderr=subprocess.PIPE)
-        if result.returncode != 0:
-             raise Exception(f"FFmpeg failed: {result.stderr.decode()}")
-        
-        # 5. ASR (Upload MP3 to tmp and transcribe)
-        logger.info("Transcribing...")
-        asr = ASRClient()
-        # Run ASR in thread
-        transcript = await asyncio.to_thread(asr.upload_and_transcribe_sync, mp3_path)
-        
-        if not transcript:
-            raise Exception("Transcription failed (returned empty)")
-
-        # 6. LLM Title Generation (Enhancement)
-        # If the title is still Unknown or weak, OR if we just want to ensure we have a good title.
-        # The user said: "Alternatively, call LlmUtil.py to summarize title".
-        # Let's do it if title is Unknown or matches default filename pattern, OR if extracted title was also missing.
-        if (not title or title == "Unknown Title" or title == "Unknown"):
-            try:
-                logger.info("Generating title from transcript via LLM...")
-                prompt = f"请根据以下视频文案总结一个简短的标题（20字以内），不要包含任何解释性文字，直接返回标题：\n\n{transcript[:1000]}"
-                
-                llm_title_chunks = []
-                # get_llm_response is already async
-                async for chunk in get_llm_response(prompt, stream=False):
-                    llm_title_chunks.append(chunk)
-                llm_title = "".join(llm_title_chunks)
-
-                if llm_title:
-                    # Clean up quotes if any
-                    llm_title = llm_title.strip().strip('"').strip('“').strip('”')
-                    logger.info(f"LLM generated title: {llm_title}")
-                    # We overwrite the title if LLM succeeds
-                    title = llm_title
-            except Exception as llm_e:
-                logger.warning(f"LLM Title generation failed: {llm_e}")
-
-        # 7. Save to DB (Update)
-        logger.info("Saving to DB...")
-        await asyncio.to_thread(update_record, request_id, title, obs_url, transcript, "COMPLETED")
-        logger.info(f"Task {request_id} completed successfully.")
-        
-    except Exception as e:
-        logger.error(f"Task {request_id} failed: {e}", exc_info=True)
-        await asyncio.to_thread(update_status, request_id, "FAILED", str(e))
-    finally:
-        # 8. Cleanup
-        if os.path.exists(temp_dir):
-            try:
-                # shutil.rmtree is sync, wrap it
-                await asyncio.to_thread(shutil.rmtree, temp_dir, ignore_errors=True)
-            except Exception as e:
-                logger.error(f"Cleanup failed: {e}")
-
-@router.post("/api/parse")
-def parse(request: ParseRequest, background_tasks: BackgroundTasks):
-    downloader = DouYinDownloader()
-    urls = downloader.extract_urls(request.text)
-    
-    if not urls:
-        # If no URLs found, try using the text as is (might be a direct link not caught by regex)
-        # But regex is quite broad. Let's just fail or try one.
-        # Let's assume text might be the URL if it's clean.
-        if request.text.startswith("http"):
-            urls = [request.text]
-        else:
-             raise HTTPException(status_code=400, detail="No valid URLs found")
-
-    created_ids = []
-    try:
-        conn = get_db_connection()
-        cursor = conn.cursor()
-        
-        for url in urls:
-            req_id = str(uuid.uuid4())
-            sql = """
-                INSERT INTO t_douyin_record (id, original_text, status, create_time)
-                VALUES (%s, %s, 'PENDING', %s)
-            """
-            cursor.execute(sql, (req_id, url, datetime.now()))
-            created_ids.append(req_id)
-            # Pass request.text (the full share text) so we can extract title from it
-            background_tasks.add_task(process_video_task, url, req_id, request.text)
-            
-        conn.commit()
-        conn.close()
-    except Exception as e:
-        raise HTTPException(status_code=500, detail=f"DB Init Error: {e}")
-    
-    return {"id": created_ids[0] if created_ids else None, "ids": created_ids, "status": "PENDING"}
-
-@router.get("/api/records")
-def get_records():
-    try:
-        conn = get_db_connection()
-        cursor = conn.cursor()
-        cursor.execute("SELECT * FROM t_douyin_record ORDER BY create_time DESC LIMIT 50")
-        records = cursor.fetchall()
-        conn.close()
-        
-        # Manually handle datetime serialization to be safe
-        for r in records:
-            if 'create_time' in r and r['create_time']:
-                r['create_time'] = r['create_time'].strftime("%Y-%m-%d %H:%M:%S")
-            if 'update_time' in r and r['update_time']:
-                r['update_time'] = r['update_time'].strftime("%Y-%m-%d %H:%M:%S")
-                
-        return records
-    except Exception as e:
-        logger.error(f"Get records error: {e}", exc_info=True)
-        return []
-
-@router.delete("/api/records/{id}")
-def delete_record(id: str):
-    try:
-        conn = get_db_connection()
-        cursor = conn.cursor()
-        cursor.execute("DELETE FROM t_douyin_record WHERE id=%s", (id,))
-        conn.commit()
-        conn.close()
-        return {"status": "deleted"}
-    except Exception as e:
-        raise HTTPException(status_code=500, detail=str(e))
-
-@router.post("/api/douyin/summary")
-async def generate_summary(request: SummaryRequest):
-    try:
-        # Fetch transcripts
-        conn = get_db_connection()
-        cursor = conn.cursor()
-        
-        if request.ids:
-            # Secure way to handle list in SQL
-            format_strings = ','.join(['%s'] * len(request.ids))
-            sql = f"SELECT video_name, transcript FROM t_douyin_record WHERE id IN ({format_strings}) AND status='COMPLETED'"
-            cursor.execute(sql, tuple(request.ids))
-        else:
-            # Default to latest 20
-            cursor.execute("SELECT video_name, transcript FROM t_douyin_record WHERE status='COMPLETED' ORDER BY create_time DESC LIMIT 20")
-        
-        records = cursor.fetchall()
-        conn.close()
-        
-        if not records:
-            # If no records, just return a simple message stream
-            async def empty_stream():
-                yield "未找到可总结的已完成记录，请先解析视频。"
-            return StreamingResponse(empty_stream(), media_type="text/event-stream")
-            
-        # Prepare text
-        full_text = ""
-        for r in records:
-            if r['transcript']:
-                full_text += f"【标题：{r['video_name']}】\n内容：{r['transcript']}\n\n"
-                
-        if not full_text:
-             async def empty_text_stream():
-                yield "记录中没有有效的文案内容。"
-             return StreamingResponse(empty_text_stream(), media_type="text/event-stream")
-
-        # Prompt
-        prompt = f"""
-        请对以下充电行业相关的视频内容进行知识精华提取。
-        要求：
-        1. 忽略无关闲聊和口语化表达；
-        2. 按条目列出核心知识点，不要长篇大论；
-        3. 保持简洁专业，只保留干货；
-        4. 返回格式为Markdown列表。
-        
-        内容如下：
-        {full_text[:15000]}
-        """
-        
-        # Limit context to avoid errors, 15000 chars is roughly safe for most models, 
-        # but if using a small model, might need less. Assuming robust model.
-        
-        return StreamingResponse(get_llm_response(prompt), media_type="text/event-stream")
-        
-    except Exception as e:
-        logger.error(f"Summary generation failed: {e}")
-        raise HTTPException(status_code=500, detail=str(e))
-
-async def recover_pending_tasks():
-    """
-    Check for tasks stuck in PENDING or PROCESSING state (due to server restart)
-    and restart them.
-    """
-    logger.info("Scanning for interrupted Douyin tasks...")
-    try:
-        # Use asyncio.to_thread for DB operation
-        def fetch_pending():
-            conn = get_db_connection()
-            cursor = conn.cursor()
-            # Select recent pending/processing tasks (limit 20 to avoid storm)
-            sql = """
-                SELECT id, original_text, status 
-                FROM t_douyin_record 
-                WHERE status IN ('PENDING', 'PROCESSING') 
-                ORDER BY create_time DESC LIMIT 20
-            """
-            cursor.execute(sql)
-            tasks = cursor.fetchall()
-            conn.close()
-            return tasks
-
-        tasks = await asyncio.to_thread(fetch_pending)
-        
-        if not tasks:
-            logger.info("No interrupted tasks found.")
-            return
-
-        logger.info(f"Found {len(tasks)} interrupted tasks. Restarting...")
-        for task in tasks:
-            req_id = task['id']
-            url = task['original_text']
-            # Restart task in background
-            # Note: We lost the original share text for title extraction, 
-            # so we pass empty string. It will use the URL or 'Unknown Title'.
-            # If LLM is enabled, it might fix the title later.
-            asyncio.create_task(process_video_task(url, req_id, share_text=""))
-            
-    except Exception as e:
+from fastapi import APIRouter, HTTPException, BackgroundTasks
+from fastapi.responses import StreamingResponse
+from pydantic import BaseModel
+
+# Import custom modules
+from Config.Config import OBS_CLOUD_PREFIX, OBS_BUCKET, OBS_TMP_PREFIX, OBS_SERVER
+from Util.DouYinDownloader import DouYinDownloader
+from Util.ObsUtil import ObsUploader
+from Util.ASRClient import ASRClient
+from Util.LlmUtil import get_llm_response
+from Model.DouYinModel import DouYinModel
+
+# Logger setup
+logger = logging.getLogger(__name__)
+
+router = APIRouter()
+
+class ParseRequest(BaseModel):
+    text: str
+
+class SummaryRequest(BaseModel):
+    ids: List[str] = []
+
+async def process_video_task(url: str, request_id: str, share_text: str = ""):
+    logger.info(f"Processing task {request_id}")
+    
+    model = DouYinModel()
+    # 1. Update status
+    await model.update_status(request_id, "PROCESSING")
+
+    temp_dir = os.path.abspath(f"temp_{request_id}")
+    try:
+        if not os.path.exists(temp_dir):
+            os.makedirs(temp_dir)
+            
+        # 2. Parse & Download
+        downloader = DouYinDownloader()
+        # url is passed directly now
+        if not url:
+            raise Exception("No valid URL found")
+            
+        logger.info(f"Downloading from {url}")
+        # Run download in thread to avoid blocking main loop
+        local_video_path, title = await asyncio.to_thread(downloader.download_video, url, temp_dir)
+        
+        # Title handling strategy:
+        # Priority 1: Extracted from share text (if available and valid)
+        # Priority 2: Extracted from video download (often "Unknown Title")
+        # Priority 3: Generated by LLM (done later)
+        
+        extracted_title = downloader.extract_title_from_text(share_text)
+        logger.info(f"Extracted title from text: {extracted_title}")
+        
+        # If we have a valid extracted title, use it. 
+        # But if we don't have a title yet (or it's Unknown), we definitely want to use extracted_title.
+        # Even if we have a title from yt-dlp, if it's just "Unknown Title", we prefer extracted one.
+        if extracted_title and extracted_title != "Unknown Title":
+            title = extracted_title
+        elif not title:
+            title = "Unknown Title"
+            
+        if not local_video_path or not os.path.exists(local_video_path):
+             raise Exception("Download failed")
+             
+        # 3. Upload Video to OBS (Long term storage)
+        logger.info("Uploading video to OBS...")
+        uploader = ObsUploader()
+        video_filename = os.path.basename(local_video_path)
+        obs_video_key = f"{OBS_CLOUD_PREFIX}/DouYin/{video_filename}"
+        
+        success, _ = await asyncio.to_thread(uploader.upload_file, obs_video_key, local_video_path, OBS_BUCKET)
+        if not success:
+            raise Exception("OBS Upload failed")
+            
+        # Construct public URL (Assuming standard OBS pattern or Config logic)
+        obs_url = f"https://{OBS_BUCKET}.{OBS_SERVER}/{obs_video_key}"
+            
+        # 4. Convert to MP3
+        logger.info("Converting to MP3...")
+        mp3_path = os.path.splitext(local_video_path)[0] + ".mp3"
+        cmd = [
+            "ffmpeg", "-y", "-i", local_video_path,
+            "-acodec", "libmp3lame", "-ar", "16000", "-ac", "1", "-q:a", "2",
+            mp3_path
+        ]
+        # Run ffmpeg in thread
+        result = await asyncio.to_thread(subprocess.run, cmd, stdout=subprocess.DEVNULL, stderr=subprocess.PIPE)
+        if result.returncode != 0:
+             raise Exception(f"FFmpeg failed: {result.stderr.decode()}")
+        
+        # 5. ASR (Upload MP3 to tmp and transcribe)
+        logger.info("Transcribing...")
+        asr = ASRClient()
+        # Run ASR in thread
+        transcript = await asyncio.to_thread(asr.upload_and_transcribe_sync, mp3_path)
+        
+        if not transcript:
+            raise Exception("Transcription failed (returned empty)")
+
+        # 6. LLM Title Generation (Enhancement)
+        # If the title is still Unknown or weak, OR if we just want to ensure we have a good title.
+        # The user said: "Alternatively, call LlmUtil.py to summarize title".
+        # Let's do it if title is Unknown or matches default filename pattern, OR if extracted title was also missing.
+        if (not title or title == "Unknown Title" or title == "Unknown"):
+            try:
+                logger.info("Generating title from transcript via LLM...")
+                prompt = f"请根据以下视频文案总结一个简短的标题（20字以内），不要包含任何解释性文字，直接返回标题：\n\n{transcript[:1000]}"
+                
+                llm_title_chunks = []
+                # get_llm_response is already async
+                async for chunk in get_llm_response(prompt, stream=False):
+                    llm_title_chunks.append(chunk)
+                llm_title = "".join(llm_title_chunks)
+
+                if llm_title:
+                    # Clean up quotes if any
+                    llm_title = llm_title.strip().strip('"').strip('“').strip('”')
+                    logger.info(f"LLM generated title: {llm_title}")
+                    # We overwrite the title if LLM succeeds
+                    title = llm_title
+            except Exception as llm_e:
+                logger.warning(f"LLM Title generation failed: {llm_e}")
+
+        # 7. Save to DB (Update)
+        logger.info("Saving to DB...")
+        await model.update_record(request_id, title, obs_url, transcript, "COMPLETED")
+        logger.info(f"Task {request_id} completed successfully.")
+        
+    except Exception as e:
+        logger.error(f"Task {request_id} failed: {e}", exc_info=True)
+        await model.update_status(request_id, "FAILED", str(e))
+    finally:
+        # 8. Cleanup
+        if os.path.exists(temp_dir):
+            try:
+                # shutil.rmtree is sync, wrap it
+                await asyncio.to_thread(shutil.rmtree, temp_dir, ignore_errors=True)
+            except Exception as e:
+                logger.error(f"Cleanup failed: {e}")
+
+@router.post("/api/parse")
+async def parse(request: ParseRequest, background_tasks: BackgroundTasks):
+    downloader = DouYinDownloader()
+    urls = downloader.extract_urls(request.text)
+    
+    if not urls:
+        # If no URLs found, try using the text as is (might be a direct link not caught by regex)
+        # But regex is quite broad. Let's just fail or try one.
+        # Let's assume text might be the URL if it's clean.
+        if request.text.startswith("http"):
+            urls = [request.text]
+        else:
+            raise HTTPException(status_code=400, detail="No valid URLs found")
+
+    created_ids = []
+    try:
+        model = DouYinModel()
+        for url in urls:
+            req_id = str(uuid.uuid4())
+            await model.insert_record(req_id, url)
+            created_ids.append(req_id)
+            # Pass request.text (the full share text) so we can extract title from it
+            background_tasks.add_task(process_video_task, url, req_id, request.text)
+            
+    except Exception as e:
+        raise HTTPException(status_code=500, detail=f"DB Init Error: {e}")
+    
+    return {"id": created_ids[0] if created_ids else None, "ids": created_ids, "status": "PENDING"}
+
+@router.get("/api/records")
+async def get_records():
+    try:
+        model = DouYinModel()
+        records = await model.get_records()
+        return records
+    except Exception as e:
+        logger.error(f"Get records error: {e}", exc_info=True)
+        return []
+
+@router.delete("/api/records/{id}")
+async def delete_record(id: str):
+    try:
+        model = DouYinModel()
+        await model.delete_record(id)
+        return {"status": "deleted"}
+    except Exception as e:
+        raise HTTPException(status_code=500, detail=str(e))
+
+@router.post("/api/douyin/summary")
+async def generate_summary(request: SummaryRequest):
+    try:
+        # Fetch transcripts
+        model = DouYinModel()
+        records = await model.get_transcripts(ids=request.ids)
+        
+        if not records:
+            # If no records, just return a simple message stream
+            async def empty_stream():
+                yield "未找到可总结的已完成记录，请先解析视频。"
+            return StreamingResponse(empty_stream(), media_type="text/event-stream")
+            
+        # Prepare text
+        full_text = ""
+        for r in records:
+            if r['transcript']:
+                full_text += f"【标题：{r['video_name']}】\n内容：{r['transcript']}\n\n"
+                
+        if not full_text:
+             async def empty_text_stream():
+                yield "记录中没有有效的文案内容。"
+             return StreamingResponse(empty_text_stream(), media_type="text/event-stream")
+
+        # Prompt
+        prompt = f"""
+        请对以下充电行业相关的视频内容进行知识精华提取。
+        要求：
+        1. 忽略无关闲聊和口语化表达；
+        2. 按条目列出核心知识点，不要长篇大论；
+        3. 保持简洁专业，只保留干货；
+        4. 返回格式为Markdown列表。
+        
+        内容如下：
+        {full_text[:15000]}
+        """
+        
+        # Limit context to avoid errors, 15000 chars is roughly safe for most models, 
+        # but if using a small model, might need less. Assuming robust model.
+        
+        return StreamingResponse(get_llm_response(prompt), media_type="text/event-stream")
+        
+    except Exception as e:
+        logger.error(f"Summary generation failed: {e}")
+        raise HTTPException(status_code=500, detail=str(e))
+
+async def recover_pending_tasks():
+    """
+    Check for tasks stuck in PENDING or PROCESSING state (due to server restart)
+    and restart them.
+    """
+    logger.info("Scanning for interrupted Douyin tasks...")
+    try:
+        model = DouYinModel()
+        tasks = await model.get_interrupted_tasks()
+        
+        if not tasks:
+            logger.info("No interrupted tasks found.")
+            return
+
+        logger.info(f"Found {len(tasks)} interrupted tasks. Restarting...")
+        for task in tasks:
+            req_id = task['id']
+            url = task['original_text']
+            # Restart task in background
+            # Note: We lost the original share text for title extraction, 
+            # so we pass empty string. It will use the URL or 'Unknown Title'.
+            # If LLM is enabled, it might fix the title later.
+            asyncio.create_task(process_video_task(url, req_id, share_text=""))
+            
+    except Exception as e:
         logger.error(f"Failed to recover tasks: {e}", exc_info=True)

Controller/HaiBaoController.py

@@ -9,7 +9,7 @@ from sqlalchemy.sql import text
 
 from Util.BananaClient import BananaClient
 from Util.LlmUtil import get_llm_response
-from DbKit.Db import Db
+from Model.HaiBaoModel import HaiBaoModel
 
 router = APIRouter(prefix="/haibao")
 logger = logging.getLogger("HaiBaoController")
@@ -21,43 +21,8 @@ class GenerateRequest(BaseModel):
 
 @router.on_event("startup")
 async def startup_event():
-    """初始化时检查并创建表"""
-    db = Db()
-    await db.init_db()
-    
-    # Doris 建表语句
-    create_table_sql = """
-    CREATE TABLE IF NOT EXISTS haibao_history (
-        id VARCHAR(50) COMMENT "ID",
-        prompt TEXT COMMENT "提示词",
-        image_url VARCHAR(500) COMMENT "图片URL",
-        scheme_content TEXT COMMENT "文案方案",
-        created_at DATETIME COMMENT "创建时间"
-    )
-    DUPLICATE KEY(id)
-    DISTRIBUTED BY HASH(id) BUCKETS 1
-    PROPERTIES (
-        "replication_num" = "1"
-    );
-    """
-    try:
-        # 使用 engine 直接执行 DDL
-        async with db.engine.begin() as conn:
-            await conn.execute(text(create_table_sql))
-            
-            # 尝试添加列（如果表已存在但列不存在）
-            # Doris 不支持 IF NOT EXISTS for ADD COLUMN directly nicely in all versions without error if exists
-            # 所以这里简单捕获异常，如果列已存在则忽略
-            try:
-                alter_sql = "ALTER TABLE haibao_history ADD COLUMN scheme_content TEXT COMMENT '文案方案'"
-                await conn.execute(text(alter_sql))
-            except Exception as e:
-                # 忽略列已存在的错误
-                pass
-                
-        logger.info("海报历史表检查/更新成功")
-    except Exception as e:
-        logger.error(f"海报历史表创建/更新失败: {e}")
+    """初始化时逻辑"""
+    logger.info("海报控制器启动成功")
 
 class RefineRequest(BaseModel):
     prompt: str
@@ -184,26 +149,11 @@ async def generate_poster(req: GenerateRequest):
         image_url, scheme_content = await asyncio.gather(generate_image_task(), generate_text_task())
         
         # 4. 保存到数据库
-        db = Db()
+        model = HaiBaoModel()
         record_id = str(uuid.uuid4())
         created_at = datetime.now()
         
-        insert_sql = """
-        INSERT INTO haibao_history (id, prompt, image_url, scheme_content, created_at)
-        VALUES (:id, :prompt, :image_url, :scheme_content, :created_at)
-        """
-        
-        params = {
-            "id": record_id,
-            "prompt": req.prompt,
-            "image_url": image_url,
-            "scheme_content": scheme_content,
-            "created_at": created_at
-        }
-        
-        async with await db.get_session() as session:
-            async with session.begin():
-                await session.execute(text(insert_sql), params)
+        await model.insert_record(record_id, req.prompt, image_url, scheme_content, created_at)
                 
         return {
             "id": record_id,
@@ -220,10 +170,9 @@ async def generate_poster(req: GenerateRequest):
 @router.get("/history")
 async def get_history():
     """获取海报生成历史"""
-    db = Db()
-    sql = "SELECT * FROM haibao_history ORDER BY created_at DESC LIMIT 50"
     try:
-        result = await db.find(sql)
+        model = HaiBaoModel()
+        result = await model.get_history(50)
         
         formatted_result = []
         for item in result:

Controller/YltAnalyticsController.py

@@ -6,7 +6,6 @@ from typing import List, Optional, Dict, Any
 from fastapi import APIRouter, HTTPException
 from fastapi.responses import StreamingResponse, FileResponse, JSONResponse
 
-from DbKit.Db import Db
 from Config.Config import DB_URL
 from Util.LlmUtil import get_llm_response
 from Tools.T6_Export import export_excel, DorisExcelExporter, extract_hourly_prices_from_schedule
@@ -24,31 +23,35 @@ from Model.YltAnalyticsModel import (
     PriceSeries,
     PriceComparisonResponse,
     PriceComparisonSummary,
+    YltAnalyticsModel,
 )
 
 
+from DbKit.Db import Db
+
 router = APIRouter()
 
-db = Db(db_url=DB_URL)
+# db = Db(db_url=DB_URL) # Removed direct db instance
 
 
 async def init_db():
+    db = Db()
     await db.init_db()
 
 
 async def close_db():
+    db = Db()
     await db.close()
 
 
 @router.get("/api/operators/hourly-prices")
 async def get_operators_hourly_prices():
     operators = ["新电途", "特来电", "驿来特", "艾特吉易充"]
-    exporter = DorisExcelExporter(db_url=DB_URL)
-    await exporter.init()
+    model = YltAnalyticsModel()
     try:
         result = []
         for op in operators:
-            rows = await exporter.fetch_current_station_rows(op)
+            rows = await model.fetch_current_station_rows(op)
             if not rows:
                 result.append({"operator": op, "series": [None] * 24})
                 continue
@@ -71,11 +74,72 @@ async def get_operators_hourly_prices():
                 else:
                     avg_series.append(None)
             result.append({"operator": op, "series": avg_series})
-    finally:
-        await exporter.close()
+    except Exception as e:
+        raise HTTPException(status_code=500, detail=str(e))
     return {"operators": result}
 
 
+@router.get("/api/operators/price-trends")
+async def get_operators_price_trends(days: int = 7):
+    operators = ["新电途", "特来电", "驿来特", "艾特吉易充"]
+    
+    model = YltAnalyticsModel()
+    rows = await model.get_operators_price_trends(days)
+    
+    # 数据结构: { operator: { date_str: [sums_of_24h, counts_of_24h] } }
+    trend_data = {}
+    for op in operators:
+        trend_data[op] = {}
+        
+    for row in rows:
+        op = row.get("operator")
+        if op not in trend_data:
+            continue
+        d_str = str(row.get("date_str"))
+        schedule_json = row.get("schedule_json")
+        
+        if d_str not in trend_data[op]:
+            trend_data[op][d_str] = {"sums": [0.0] * 24, "counts": [0] * 24}
+            
+        series = extract_hourly_prices_from_schedule(schedule_json)
+        for i in range(24):
+            v = series[i]
+            if v is not None:
+                trend_data[op][d_str]["sums"][i] += float(v)
+                trend_data[op][d_str]["counts"][i] += 1
+                
+    # 转换为 ECharts 友好格式
+    # 1. 获取所有日期并排序
+    all_dates = sorted(list(set(str(row.get("date_str")) for row in rows)))
+    
+    # 2. 为每个运营商计算每天的平均价格（24小时的平均值）
+    series_result = []
+    for op in operators:
+        op_trend = []
+        for d in all_dates:
+            if d in trend_data[op]:
+                day_stats = trend_data[op][d]
+                day_avg_sum = 0.0
+                day_hour_count = 0
+                for i in range(24):
+                    if day_stats["counts"][i] > 0:
+                        day_avg_sum += (day_stats["sums"][i] / day_stats["counts"][i])
+                        day_hour_count += 1
+                
+                if day_hour_count > 0:
+                    op_trend.append(round(day_avg_sum / day_hour_count, 4))
+                else:
+                    op_trend.append(None)
+            else:
+                op_trend.append(None)
+        series_result.append({"name": op, "data": op_trend})
+        
+    return {
+        "dates": all_dates,
+        "series": series_result
+    }
+
+
 @router.get("/api/export/prices-zip")
 async def export_prices_zip():
     operators = ["新电途", "特来电", "驿来特", "艾特吉易充"]
@@ -177,26 +241,8 @@ async def ai_pricing_strategy_summary():
 
 @router.get("/api/ylt/stations", response_model=List[StationBase])
 async def list_ylt_stations(q: Optional[str] = None):
-    base_sql = """
-    SELECT
-        p.station_hash,
-        p.operator,
-        p.station_name,
-        p.address,
-        p.coord_x,
-        p.coord_y,
-        s.current_price
-    FROM t_station_profile_scd p
-    LEFT JOIN t_station_status_scd s
-        ON p.station_hash = s.station_hash AND s.is_current = 1
-    WHERE p.operator = '驿来特'
-    """
-    params: Dict[str, Any] = {}
-    if q:
-        base_sql += " AND (p.station_name LIKE :kw OR p.address LIKE :kw)"
-        params["kw"] = f"%{q}%"
-    base_sql += " ORDER BY p.station_name"
-    rows = await db.find(base_sql, params)
+    model = YltAnalyticsModel()
+    rows = await model.list_ylt_stations(q)
     result: List[StationBase] = []
     for r in rows:
         result.append(
@@ -225,21 +271,8 @@ def haversine_km(lon1: float, lat1: float, lon2: float, lat2: float) -> float:
 
 
 async def fetch_current_stations() -> List[dict]:
-    sql = """
-    SELECT
-        p.station_hash,
-        p.operator,
-        p.station_name,
-        p.address,
-        p.coord_x,
-        p.coord_y,
-        s.current_price
-    FROM t_station_profile_scd p
-    LEFT JOIN t_station_status_scd s
-        ON p.station_hash = s.station_hash AND s.is_current = 1
-    WHERE p.is_current = 1
-    """
-    return await db.find(sql, {})
+    model = YltAnalyticsModel()
+    return await model.fetch_current_stations()
 
 
 async def build_geo_competition(station_hash: str, radius_km: float = 3.0) -> GeoCompetitionResponse:
@@ -318,17 +351,8 @@ async def build_geo_competition(station_hash: str, radius_km: float = 3.0) -> Ge
 
 
 async def fetch_station_schedule_json(station_hash: str) -> Optional[str]:
-    sql = """
-    SELECT schedule_json
-    FROM t_station_price_schedule_scd
-    WHERE station_hash = :h AND is_current = 1
-    ORDER BY valid_start_time DESC
-    LIMIT 1
-    """
-    rows = await db.find(sql, {"h": station_hash})
-    if not rows:
-        return None
-    value = rows[0].get("schedule_json")
+    model = YltAnalyticsModel()
+    value = await model.fetch_station_schedule_json(station_hash)
     if value is None:
         return None
     if isinstance(value, str):